Generative AI and Innovation

Grok 3 has just been released, and by several metrics it’s the best LLM out there. I wrote a couple posts today that related to Grok 3.

The first was commenting on an article by Alberto Romeo on Grok 3 and what Rich Sutton calls “the bitter lesson.” I asked Grok 3 to summarize what exactly the “bitter lesson is” and it concluded with the following:

In essence, the Bitter Lesson is: computation trumps human-crafted specialization in AI, and embracing this, though humbling, is key to future success.

The other post related to Grok 3 revisited using generative AI to create mnemonic images. I wrote about this in 2022, and the results at the time were disappointing to say the least. I tried the same examples using Grok 3 and the results were far better.

Here is an image of Grover holding an onion and a wiener dog. To find out why on earth anyone would possibly want such an image, see the post.

Wayne also posted an article today. Usually he’s the one who writes about AI, but today we switched places. His new post looks at the role of institutions in innovation. He discusses the book The Sociology of Philosophies which argues that truly innovative ideas come out of institutions of innovators. Wayne contrasts this with the idea of maverick science which asserts the opposite, the breakthrough ideas come from outsiders.